Automation, Productivity, and Human Work
What to make of automation, and what automation will make of us, are two of the most important questions being asked by academics, scientists, politicians, and the public. Discussion about automation are with plenty of support, scepticism, predictions made, and questions to be answered. I will discuss and summarize a few news articles on automation that have made the rounds on the Internet. Then, how the ideas discussed fit into our project looking at automation in healthcare. As I have met many primary care doctors across England, I would like to share some thoughts on how automation might impact their work.
To start, we shall look at a worrisome phenomenon occurring in the background of modern economies: stagnation of productivity. Why is this worrisome? Technological progress always increases productivity and as productivity rises we see higher wages and living standards. This relationship in the U.S. has typically been 1:1. However, over the past 20 years we have seen wages fall behind productivity. More troublesome is that new data from the U.S. Bureau of Labor Statistics shows disappointing growth since 2007 and even more so since 2010. A productivity slowdown is also seen in the economies of Japan, Germany, France, and the United Kingdom. A recent article in the New Yorker called The Great Productivity Puzzle submits three theories on why this is happening. To summarize:
- This phenomenon is a sticky wicket to sort out statistically. It may not be measured correctly or accurately reflect GDP and household incomes. Particularly in IT sectors and new areas of online labour.
- The productivity slowdown is a result of the great recession. Businesses entered a mode in which they make do with the capital goods they have without making any new capital investments.
- Technologies do not make the impact they once did. Innovations like electricity, the Internet, indoor plumbing, and shipping containers were truly transformative. To have the Internet in our pocket via smartphone is amazing, but it cannot compare with how essential clean running water is in your home.
The Great Productivity Puzzle article concludes that the slowdown in productivity is likely a mix of all three points. The interesting question is: where do we go from here? Enter big data, artificial intelligence, robotics, automation, and other advanced technologies. Will these technologies go the route of theory three? Or, will they present a meaningful shift in economic productivity?
Mike Osborne and Carl Frey published a paper in 2013 that looked at 702 occupations’ probability for automation. They estimate about 47% of total US labour force is at risk for having jobs replaced by automated labour. The variables that indicated a negative relationship with automation? These are income and educational attainment. The greater income you collect and higher education you have, it becomes harder to automate your job. Additionally, if you look at the top 25 occupations that present the greatest resistance to automation, nearly all of them are in healthcare related occupations. More on that topic later.
One of the first professions that will undergo rapid transformation over the next 5 to 10 years is the long haul truck driver. This change occurs from strides in autonomous driving technologies seen in the Google car, Otto, Tesla, and many global car manufacturers researching their own driverless car technology. Truck driving is an interesting profession to look at because it is one of the few remaining jobs that pays a decent middle class income and does not require a college degree. In 2014, the U.S. Bureau of Labor Statistics noted about 1.8 million heavy truck drivers. That’s a large labour force to disrupt. A recent Vox article remarks about autonomous trucks and advocates for a universal basic income for those careers affected.
This gets us into a discussion about the meaning and fulfilment of work. How do you feel when you have a satisfying career you have worked over 10 years at that has supported you and your family only to be paid to not do it or, worse, only perform the least exciting part of your old job? This was the case in a New York Times interview where a truck driver for over 10 years stated there is “not a chance in hell” that a driver would sleep while the truck drives itself. Or, if you are paying the human to only drive the last few miles into a destination that are typically tricky and difficult for computers to navigate, that same driver said it would “not make the job worth doing”.
The first go to “solution”, the one everyone talks about, is to retrain people. It’s far more challenging than it sounds. How do you retrain so many people who have put their lives into a profession for decades? Furthermore, what do you retrain them to do? The occupation type that is being overtaken by automation and the amount of time a given worker has invested in that work will combine to create this challenge. Not everyone can retrain and not everyone may want to retrain. People sometimes have deep relationships with their occupations; a job can be part of one’s personality and identity.
However, what if the human labour for some professions resists automation simply because it is seen as artisanal, hand crafted, bespoke, and of added value. This is the thought behind another Vox article which argues that in a highly automated industry human labour becomes a luxury good. Archetypal of this labour is the third wave coffee movement. There is purposefully as little automation as possible. The entire coffee production line is carefully done by hand to produce the highest quality coffee possible. A local coffee shop that sourced single origin beans then roasted those beans themselves would turn heads and produce a spit take from discerning coffee drinkers if they then used an automated coffee machine to prepare drinks. The same Vox article notes that Starbucks had mechanized the coffee process, and baristas were informed to slow down and only make one drink at a time. While there is probably not a demand for artisanal, hand crafted human driving, the article raises a point about some of the unintended consequences and different kinds of thinking that automation may force us to produce.
There is an argument that the Information Technology sector has created more jobs than it has eliminated. Data over 140 years shows that technology has been a job creator rather than a terminator of human labour. Since the 1800s, there has been an ebb and flow of new technologies that augment or ultimately render human work obsolete only to give way to new areas of work. A quick Google search for jobs that no longer exist produces a litany of extinct professions. Non-existent job titles that are in recent memory such as milkman and switchboard operators to lamp lighters, human alarm clocks, pre-radar listeners, ice cutters, bowling alley pinsetters, and lectors who read aloud the news and literature to factory workers. All of the aforementioned jobs were replaced by various technologies. Is the economy better off having eliminated all those jobs? Undoubtedly. We also have the benefit of several decades and in some cases hundreds of years to look back and situate those job losses in the greater historical context. Will we look back at humans that once drove cars and trucks themselves and think: how silly? When smartphones were first popularized, who could have predicted the rise of the mobile software developer, digital nomads, the rise of the software platform, and Silicon Valley start-ups based on a phone application?
So far, we have looked at an occupation under threat from automation, and presented the idea that human labour as a luxury good may still prosper in certain sectors in a highly automated world. We will see labour and job titles shift as new technologies diffuse throughout economic sectors. What I think will be of great importance for the future of advanced economies is understanding how humans and machines may work together or how advanced technologies will continue to augment human labour. We already see new emerging human-computer relationships in the form of chat bots. Humans text utilizing narrow artificial intelligence, algorithms, and scripts which use natural language to gain information. Nineteen-year-old Joshua Browder created a robot lawyer to help people appeal parking fines. Entering a text conversation with this robot, as you would instant message with a colleague or friend, helps people determine the likelihood that their appeal is successful and guides them through the process of the appeal. This is one example of humans interacting with non-human entities. There are others in the services sector such as Fixmystreet or corporate implementations such as Microsoft, Apple, Facebook, and Google employing these technologies into their respective platforms.
Text is one way to interact with a robot. But what about work in technologically dense environments that require highly skilled human work? Remember the introduction to Osborne and Frey’s paper in the beginning of this blog post? That nearly all of the occupations resistant to automation are in the healthcare sector? This is an exciting area to look at automation for three reasons. First, there is the challenge of automation in this domain. Healthcare occupations contain a heavy dose of bottlenecks to computerization. Skills that require creative intelligence are a challenge for computers to reproduce. Think creative problem solving, coming up with unusual ideas or counter intuitive methods to solve problems. Thinking fast on your feet given the context and situation is something humans are good at; machines, not yet. Another challenge is social intelligence. Every day we navigate the world though our skills of persuasion, negotiation, empathy, read body language of those we converse, understand the tension of a situation, and use humour to influence a situation. Social intelligence is an important part of the work of healthcare professionals. At present, social intelligence is a wicked hard human attribute to imbue into technology. While these skills may be challenges to both automation and computerisation in the healthcare sector, they also represent opportunities to apply some creative thinking and social intelligence to explore ways that technologies may not necessarily automate social and creative skills, but augment those skills for humans.
The second exciting thing about automation in primary care is the potentially transformative impact the appropriate application of automation technologies can have in NHS. With little effort, it is easy to read about NHS funding issues, increased workloads for medical staff, and more patients requesting appointments. One path toward providing some relief is for people to only focus on the tasks that are important to them and that they (hopefully) enjoy doing. Automation is one way to attack that problem. What if we could give people back hours and hours of time by eliminating the need for them to perform certain tasks? It’s an exciting prospect to think about: understanding the work practices of primary care employees and how to best support their work and to have a positive impact on work style. Automation already exists in primary care in certain tasks. Generating letters, patient check-in, calling patients to their appointment from the waiting room, these are all tasks that involve various levels of automation. There are still many opportunities for further understanding and implementing automation in primary care, leading to my third point.
The final exciting outlook about automation in primary care comes from meetings and informal conversations where I have engaged with general practitioners. Every general practitioner I have had a discussion with has pointed out aspects of work they perform which they would be happy to hand over to automation. I get the sense that much of this is paperwork and other redundancies that build up in organizational practice. One important application of automation would be using it to prevent things like re-work. These are instances where a clinician or staff is interrupted during a task and upon returning to the task part of the work may need to be repeated. Or, if the task is delegated and it’s unclear if it was properly finished, a portion of the work may be repeated as a safety check. These scenarios waste work and time. Looking for these instances of re-work and applying technology to close that gap is an invigorating application that can save time and money for primary care surgeries.
Over time, we will see more automation encroach into our lives and our work. Unless a job is fully automated, like we are seeing with the long distance truck driver, the work of the future will involve humans and technologies working side by side for some time. In thinking about automation of occupations like healthcare where human and technology collaborations are becoming commonplace, designers and developers must consider workers’ knowledge and experience. It is not the understanding of how to automate a particular task, but the value placed on that task that makes it worth doing. Greek poet Hesiod articulated the vital role work plays to the human experience some three thousand years ago:
“But you must work, Perses; Gods and men disapprove of that man who lives without working; Work is no disgrace, it is idleness which is a disgrace; And whatever be your lot, work is best for you; Work, work upon work!”.
What I am interested in through the automation of primary care work is not only the automation of work but the design of workflow and workloads, automating the boring bits and leaving the fulfilling work to humans. There is an idea that a world without work is coming. While that may be true, I suspect that some people actually like their profession. Maybe the answer is not to automate all work and eliminate all human work, but to allow ourselves to perform the work that fulfils us.